What is the Marketing Super Agent?

It’s an AI marketer built into Treasure Data AI Marketing Cloud. It helps marketers brainstorm ideas, develop strategy, plan campaigns, and orchestrate execution across channels. On its own, it brings clarity and speed to complex marketing decisions. Connected to Treasure Data, it becomes something more powerful: an intelligent system that understands your customers, your business, and your goals, and turns strategy into action. It combines strategic intelligence with operational execution across AI Marketing Cloud.

Do I need Treasure Data to use it?

Marketing Super Agent can provide value on its own for brainstorming and planning. When connected to Treasure Data, it becomes significantly more powerful by leveraging unified first-party customer data and activation capabilities.

How does it use customer data?

When connected to Treasure Data Intelligent Customer Data Platform (CDP), the agent uses unified customer profiles and behaviors to inform strategy, recommend audiences, and personalize journeys and messaging.

Can it execute campaigns or just plan them?

It does both. Marketing Super Agent helps define strategy and then works directly with Treasure Data Intelligent CDP and AI Marketing Cloud to build audiences, design journeys, generate content, and activate campaigns.

Is this just another chatbot?

No. Marketing Super Agent is purpose-built for marketing teams, deeply integrated with data and execution systems. It is designed to collaborate with marketers, not just answer questions.

How does it improve over time?

The agent continuously learns from your data, campaigns, and outcomes. As it gains more context, its recommendations become more accurate, relevant, and impactful.

Marketing Super Agent is built for modern marketing teams that want to move faster, think smarter, and execute more effectively using AI grounded in trusted first-party data.

Treasure Code: AI-native Data Intelligence

Conversational by default, transform customer data operations with a unified, AI-native interface. Automate workflows, manage customer data platform (CDP) configurations, and design customer journeys. All with the simplicity of natural-language commands.

This is data intelligence reimagined.

Natural language execution

Turn technical intent into production reality. Command data, segments, and workflows using plain language. Eliminate the need for complex SQL or CLI syntax. Build at the speed of thought.

Code-grade governance

Manage CDP configurations as version-controlled code with peer reviews and instant rollbacks.

Zero-friction operations

Consolidate fragmented consoles and scripts into one consistent command layer that automates deployments from development to production in seconds.

“At Thomson Reuters, our focus is on turning ideas into real outcomes for customers. By combining Treasure Data’s real‑time customer data platform and AI Agent Foundry with Treasure Code, we allow our teams to define the audiences and experiences that matter—while the platform handles more of the complexity. The result is more time spent delivering relevant, real‑time customer experiences, at scale.”

Anand Shah Director, Experience Platforms, Thomson Reuters

What's possible with Treasure Code

Deploy your entire CDP as versioned code

Bring software engineering rigor to data operations. With Treasure Code, you can manage your CDP configurations just like a software application. This means you can peer-review changes, track versions, and trigger instant rollbacks, ensuring your production environment remains stable and secure.

Evolve your segments via conversation

Update your data strategy at the speed of thought. Need a new derived attribute for your "Parent Segment"? Just ask. Instead of digging through SQL layers, you can command the platform to flag profiles or create new attributes—like identifying existing customers—using natural language that Treasure Code translates into production-ready code.

Build complex reports in minutes, not months

Turn months of manual logic into instant insights. What used to take an entire quarter of engineering—like building a comprehensive waterfall report with integrated segment creation—now happens in minutes. Simply describe the report you need, and Treasure Code handles the heavy lifting, from data architecture to the final segment output.

Access an "expert-in-the-box" for Treasure Data syntax

Skip the documentation hunt and code like a pro. Treasure Code comes with native "Treasure Data Skills," meaning it already knows the intricacies of Presto, Hive, and Digdag. Whether refactoring a pipeline or troubleshooting a user-defined field, you have a built-in platform expert that applies best practices to every command you give.

“With Treasure Code, it feels like we’ve added a legion of data engineers to the team. I can describe what I want in plain language and get production‑ready SQL, segments, and workflows in minutes.”

Tomohiko Sugiura Executive Vice President, Dentsu Digital

Classic vs AI-native experience

Capability	The classic experience	The AI-native experience
Interface	Navigating specialized UI consoles and menus.	A unified, conversational "command brain."
Logic creation	Hand-coding SQL, Hive, and Presto queries.	Natural language intent translated to code instantly.
Complexity	2-month cycles for advanced reports (e.g., Waterfalls).	Minutes to build, from logic to segment creation.
Deployment	"Click-ops" manual configuration and setup.	Code-grade deployments with version control.
Onboarding	Learning deep platform architecture over weeks.	Built-in skills guide users through every step.
Documentation	Manual wiki updates and tribal knowledge.	Self-documenting pipelines generated automatically.
Troubleshooting	Diving into logs and scripts to find errors.	Asking the built-in agent to identify and fix logic bugs.
Maintenance	Managing fragmented scripts across the org.	One consistent command layer for the entire platform.

“Treasure Code is a transformative feature that fundamentally changes how we implement and operate our CDP. Historically, constructing data pipelines and workflows required highly specialized technical skills. With Treasure Code, however, we can now execute these tasks rapidly and easily using natural language. This allows even non-experts to maintain high-quality implementations, and we are eager to actively leverage this capability to accelerate our business.”

Hideo Ota Senior Vice President & Executive Officer, NTT Integration

FAQs

Is Treasure Code a replacement for the Treasure Data UI?

Not exactly. Think of it as a "high-performance overlay." While you can still use the console for visual checks, Treasure Code provides a unified command layer that allows you to execute operations, write code, and manage workflows much faster than clicking through multiple menus.

How do I know the AI is generating the correct SQL or configuration?

Safety is built-in. Treasure Code follows a "human-in-the-loop" model where natural language commands are translated into code for your review before execution. Because it treats configurations as code, you can use peer reviews and version control to ensure everything meets your enterprise standards.

Do I need to be a developer or a "prompt engineer" to use this?

No. The goal of Treasure Code is to lower the technical floor. If you can describe your data goal—like "I need to flag users who haven't purchased in 30 days"—the built-in Claude Agent and TD Skills handle the underlying syntax of Presto, Hive, or Digdag for you. We understand not everyone is comfortable working in a command-line interface, so we’re building an application to make these capabilities accessible to non-technical users as well – stay tuned.

How do "self-documenting pipelines" actually work?

As you build or update workflows via Treasure Code, the AI analyzes the logic and generates a natural language summary of the deployment. This creates a living record of your data architecture, making it easy for anyone to understand the "why" behind a segment without hunting through scripts.

Can I use this with my existing Treasure Data setup?

Yes. Treasure Code is designed to sit on top of your current Treasure Data environment. It can "read" your existing segments and workflows, allowing you to manage legacy configurations using the new AI-native command layer immediately.

Does the AI have access to my actual customer PII?

No. The AI agent interacts with the metadata and the structural logic of your CDP (the "command brain") to build segments and workflows. Your actual customer data remains securely within the Treasure Data production environment, governed by your existing privacy and security policies.